Image-Based RNA Interference Screening Identifies Microenvironmental Signals Supporting Primary Acute Lymphoblastic Leukemia Cell Survival.

Abstract 2348

Leukemia cells are critically dependent upon interactions with the microenvironment in the bone marrow and at extramedullary sanctuary sites, which is likely to provide a protective mechanism to escape chemotherapy. In vitro, co-culture of primary ALL cells on human bone marrow derived mesenchymal stromal cells (MSCs) provides survival cues allowing long-term cultures. In contrast, primary ALL cells rapidly undergo cell death when cultured without stromal support. We developed an automated microscopy-based approach to identify pro-survival signals by RNA interference of candidate genes in MSCs and subsequent evaluation of leukemia cell survival, enabling us to functionally profile primary leukemia cells. We took advantage of our leukemia xenograft system as a renewable source of well characterized samples derived from cases with very high risk (VHR) ALL, which were selected based on clinical resistance to chemotherapy. Based on gene expression and cell surface proteomic data that we had obtained from both cellular compartments, we generated a customized siRNA library for 110 candidate genes with a potential function in stromal support. Primary ALL cells were seeded on reversely transfected MSC cells and ALL cell viability was assessed with a fluorescent vital dye after 6 days. Image analysis and machine learning algorithms were developed for quantification of surviving ALL cells on top of MSC.

Evaluating three VHR-ALL cases we observed a strong decrease of viability when interfering with the expression of 14 candidate genes in 2 out of 3 patients. Interestingly, in validation studies with 7 additional cases, the pattern of dependence on the genes tested were confirmed to be only partially overlapping between patients, indicating the existence of functional differences in distinct subsets. As a proof of concept, we could show that down-regulation of VCAM1 or the VEGF pathway in MSCs decreased ALL survival supporting earlier studies. Concomitantly, inhibitors of VEGF signalling recapitulated ALL cell viability decrease for patients that showed to be dependent on VEGFC expression in MSCs. One of the strongest effects on ALL survival was achieved by down-regulation of the membrane protein Basigin (CD147). Specifically, 13 out of 17 ALL cases were affected by the modulation of Basigin on MSC level. Basigin has been implicated in cell signalling, in interactions with extracellular matrix and serves as chaperone to different membrane carrier proteins. Among putative Basigin interactors we identified the heteromeric amino acid transporter SLC3A2 (CD98) to be required for ALL survival in the same set of ALL cases. The down-regulation of Basigin, SLC3A2 or both together in MSC cells induces an increase in ROS in ALL cells resulting in apoptotic cell death, which indicates that Basigin/SLC3A2 function is important for the integrity of ALL cell metabolism in this model of the leukemia niche. We are now investigating which metabolites are implicated in the mechanism of action.

Taken together, we have established a robust platform for systematic functional investigation of primary ALL survival in a 2-D model of the microenvironment and obtained evidence for patient-specific dependence of leukemia cell survival on stromal support. Critical interactions between ALL cells and bone marrow stromal cells can be identified with this approach, which will be useful for unbiased higher throughput screening and combinatorial testing. This platform will also be of great interest for preclinical drug profiling on clinically relevant patients samples in the context of protective bone marrow signals.